Trophic status strongly regulates nitrous oxide but not methane production in global freshwater lake sediments

Freshwater lakes are vital ecosystems, providing essential services such as drinking water and irrigation. However, their position in the landscape makes them highly susceptible to terrestrial nutrient inputs, and their long water retention times facilitate the accumulation of pollutants, including excess nitrogen from anthropogenic sources¹. The worldwide application of industrial nitrogen fertilizers has greatly accelerated the global nitrogen cycle, leading to excessive leaching into aquatic ecosystems^2,3,4. According to the United Nations Food and Agriculture Organization (FAO), global fertilizer demand is projected to increase by 50% by 2050^5,6, further exacerbating nitrogen pollution, a key driver influencing the ecological sustainability of aquatic ecosystems⁷. Elevated nitrogen loading stimulates biological processes that promote the proliferation of primary producers and fuel harmful algal blooms⁸. This phenomenon, known as eutrophication, results in an ecological state marked by algal proliferation, hypoxia, anoxia, and biodiversity loss⁹.

Despite these pressures, freshwater ecosystems serve as essential hotspots for the removal of excessive anthropogenic nitrogen. Lakes alone remove up to 90% of watershed-derived nitrogen¹⁰ and contribute approximately one-third of global land-derived nitrogen elimination¹¹. These removal processes are largely driven by microbial nitrogen cycling, including nitrification, denitrification, and anammox, with the latter two serving as the primary pathways for permanent nitrogen loss^12,13,14. Although anammox and denitrification hotspots frequently co-occur, quantitative assessments show that denitrification removes nitrogen at a rate approximately eight times higher than anammox in inland aquatic ecosystems¹⁵. Global lake denitrification alone was estimated at 19–43 Tg N yr⁻¹, accounting for 7–16% of terrestrial nitrogen inputs¹¹. The efficiency of nitrogen removal via denitrification depends heavily on environmental conditions. Eutrophic lakes, for instance, remove twice the areal nitrogen load of oligotrophic lakes, largely due to abundant organic carbon that supports denitrifying bacteria¹³.

Although lakes provide the crucial ecosystem service of removing excess anthropogenic nitrogen, eutrophication has compromised this function by stimulating the production of potent greenhouse gases (GHGs), notably methane (CH₄)¹⁶ and nitrous oxide (N₂O)^17,18. Lakes contribute approximately 18.6% and 0.5% of global CH₄ and N₂O emissions, respectively^19,20. Nitrification and denitrification are the primary N₂O production pathways^21,22. This involves ammonia-oxidizing bacteria (AOB) generating N₂O via nitrifier denitrification (NO₂⁻ → N₂O) or the oxidation of hydroxylamine (NH₂OH)^21,23,24, while denitrifiers produce N₂O via incomplete denitrification²⁵. The tight coupling of nitrification and denitrification, the primary pathway for nitrogen removal in lakes, occurs mainly in the oxic–anoxic transition zone of the upper sediments^13,26. In this zone, oxygen availability enhances nitrogen removal by promoting organic matter mineralization and nitrification, which supplies the nitrate (NO₃⁻) necessary for denitrification. Primarily as a result of increasing nitrogen loading, global N₂O emissions from lakes reached 64.6  ±  12.1 Gg N₂O-N yr⁻¹ in the 2010s, a value that has increased by 126% since the pre-industrial era²⁷. The existing literature presents a paradox regarding the drivers of lacustrine N₂O emissions. Some studies report that denitrification contributes increasingly to N₂O production with elevated eutrophication^17,18, while others identify nitrification as the dominant source in nutrient-rich lakes²⁸. This discrepancy has led to conflicting views and an uncertain mechanistic understanding of how trophic status governs N₂O fluxes. Notably, despite covering only 2% of Earth’s land surface, lake sediments constitute a disproportionally significant organic matter reservoir²⁹, due to rising nutrient inputs in recent decades³⁰. The mineralization of this organic matter, accompanied by methanogenesis performed by anaerobic archaea encoding methyl–coenzyme M reductase (MCR) complex, represents the largest biogenic source of CH₄ on Earth³¹. CH₄ emissions from global lakes were estimated at 41.6 ± 18.3 Tg CH₄ yr⁻¹³², with projected increases of 30–90% this century¹⁶.

Under fossil-fueled development, eutrophication is projected to worsen in 91% of lakes by 2050, whereas sustainable practices could reduce it in 63% of lakes³³. Consequently, future lakes worldwide will diverge along both eutrophication and oligotrophication trajectories, fundamentally shaped by regional development pathways (fossil-fueled vs. sustainable). A deeper understanding of the regulatory role of these trophic state transitions in governing CH₄ and N₂O fluxes is required to make effective GHGs management decisions, yet the underlying mechanisms remain poorly understood. To address this, we integrate metagenomics with batch cultivation to unravel the driving mechanisms of CH₄ and N₂O emissions from sediments of eutrophic and oligotrophic lakes. We expand these insights through a global metagenomic survey, establishing a framework that links their production potentials to trophic status. Finally, using an innovative cross-inoculation experiments, we simulate how eutrophication and oligotrophication processes alter sedimentary CH₄ and N₂O (nitrification vs. denitrification) emissions. Our findings provide a mechanistic understanding of trophic status-regulated CH₄ and N₂O productions, offering a science-based foundation for mitigating their ecological impacts.

Site-scale analysis of nitrification-denitrification dominance in sedimentary nitrogen removal

Lake Donghu (DH) exhibits pronounced redox stratification, with hypoxic sediments (dissolved oxygen (DO): 0.21–2.62 mg/kg) contrasting sharply with oxic water columns (DO: 8.46–10.07 mg/L) (Table S1, Fig. S1). The pH of both the water and sediments was slightly alkaline (pH 7.48 − 8.59). Sediments were rich in organic carbon (28.34–48.82 mg/g total carbon, 533.87–756.34 mg/kg soluble organic carbon), serving as a major carbon sink. High but variable inorganic nitrogen concentrations (ammonium (NH₄⁺), 15.7–66.82; NO₂⁻, 2.65–12.43; NO₃⁻, 132.11–412.20 mg-N/kg) reflected high levels of eutrophic conditions (Table S1). Metagenomic sequencing yielded 1,120 (water) and 965 (sediment) million trimmed reads, assembling into 368 and 400 metagenome-assembled genomes (MAGs) (≥ 50% completeness, ≤ 50% contamination), respectively (Table S2, Supplementary Data 1). Phylogenetic analyses of these recovered MAGs revealed distinct microbial community compositions in water and sediments (Fig. S2).

The oxic-anoxic interface of lake sediments supported active nitrogen cycling, fueled by settling organic matter from the water column¹¹. Both nitrogen fixation (nifH) and organic nitrogen mineralization (ureC) were prominent in water columns and sediments (Figs. 1, S3, S4, S5b, e). The rapid ammonium assimilation likely maintained low dissolved inorganic nitrogen concentrations in surface waters (Table S1). This pattern aligns with observations in other freshwater systems where biological demand rapidly depletes available nitrogen³⁴. While nitrifying genes, including ammonia monooxygenase (amo), hydroxylamine oxidoreductase gene (hao), and nitrite oxidoreductase (nxr), were weakly detected in oxic water columns (Fig. S7), their substantial presence in sediments (Figs. 1, S8) indicated robust nitrification potentials. The coupling of organic matter oxidation and nitrification activity in sediments likely promoted localized anoxia, creating favorable conditions for nitrogen removal through denitrification and anammox.

Notably, anammox-related genes were not detected in any water and sediment samples, while denitrification genes were readily identified with relatively high abundance in sediments (Figs. 1, S8). This finding was consistent across our global metagenomic survey of freshwater lake sediments, where core anammox genes, hydrazine synthase (hzs) and hydrazine dehydrogenase (hdh) coding genes, were consistently absent (data not shown). ¹⁵N isotope tracing experiments provide conclusive evidence supporting these genomic observations, demonstrating that denitrification was the predominant nitrogen removal pathway in sediments (Fig. 2a). This aligns with previous reports showing anammox contributes minimally (~8-fold less than denitrification) to nitrogen removal in freshwater ecosystems¹⁵. Significant positive correlations between denitrifying (nirK and nosZ) and nitrifying (ammonia-oxidizing archaea (AOA) amoA) genes (p < 0.05) as well as denitrifying nirS and nitrifying nxrA (p < 0.05) (Fig. S6). Positive correlations were also observed between the nitrification process and the denitrifying gene nirS (p < 0.05) as well as the denitrification process and the nitrifying gene nxrA (p < 0.01). These results indicate the tight interaction between nitrification and denitrification processes in sediments, as evidenced by isotope ¹⁵N tracing evidence¹³.

Site-scale analysis of coupled methanogenesis and nitrogen fixation in lake sediments

Nitrogen fixation is a crucial factor that influencing nitrogen availability in natural ecosystems³⁵, catalyzed by a conserved nitrogenase (NifH) in both bacteria and archaea³⁶. Our phylogenetic analysis revealed that most recovered NifH sequences from water samples were assigned to Cyanobacteria (Figs. S3, S5b). Conversely, cyanobacterial nifH genes were not recovered from sediment metagenomic datasets, and only low-abundance signals were detected through read mapping (Fig. S3). However, we recovered a high abundance of nifH genes from methanogens, with some detected at low level in water samples (Figs. S3, S5b). Diazotrophic methanogens³⁷, which represent the only archaea known to contain the nifH gene for nitrogen fixation³⁸, have been identified as key contributors to nitrogen fixation in freshwater wetlands³⁹. Therefore, the observed dominance of archaeal nifH in lake sediments implies a close coupling between nitrogen fixation and methanogenesis. Although methanogenic archaea can adapt to microoxic niches within floating cyanobacterial aggregates⁴⁰, we recovered no methanogenic MAG from water samples (Fig. S5b). Only one methanogenic MAG (bin.258) recovered from sediments was faintly detected in water through reads mapping (Fig. S2a). Methanogenic activity was completely suppressed at DO concentrations above 3.0 mg/L⁴¹, and while some taxa tolerate low oxygen levels, they remain active only under highly reduced conditions^42,43. The high DO levels of water columns (> 7 mg/L; Table S1) explain the absence of methanogens, whereas hypoxic sediments (DO, 0.21 − 2.62 mg/kg; Table S1) provide a favorable low-oxygen environment. Although methanogenic nitrogen fixation occurs in various hypoxic ecosystems (e.g., hydrothermal vents⁴⁴, deep-sea methane seeps⁴⁵, and wetlands³⁹), its role in freshwater lakes, globally important hotspots of methanogenesis⁴⁶, has been overlooked.

Freshwater lake sediments, primarily anoxic zones, are major CH₄ sources, contributing ~70% of freshwater emission⁴⁷. We detected substantial prevalence of methyl-coenzyme M reductase gene (mcrA) gene (methanogenesis marker gene) in sediments (Figs. 1, S5c), which showed a strong positive correlation with abundant nifH genes (p < 0.05; Fig. S6), suggesting a potential functional linkage between methanogenic and nitrogen-fixing. Notably, sediments with nitrogen gas (N₂) headspace exhibited significantly higher CH₄ production (0.586 ± 0.054 nmol CH₄/g_(sediment)/h) than those with argon gas (Ar) (0.285 ± 0.009 nmol CH₄/g_(sediment)/h; Fig. 3a, b), as evidenced by a linear regression model (p < 0.001; Fig. 3c). In addition, a significant higher NH₄⁺ production rate was detected in the sediment with N₂ as the headspace compared to that with Ar (2.86 ± 0.05 vs. 0.17 ± 0.05 μmol NH₄⁺/g_(sediment)/h; p < 0.001; Fig. 3d). These results demonstrate that N₂ likely stimulates methanogenesis by enhancing archaeal N₂ fixation, providing biochemical evidence for the coupling of these two processes in lake sediments, as suggested by metagenomic analysis. Similar coupling has been proposed in pond sediments, where nifH abundance highly correlated with CH₄ concentrations⁴⁸. Consistent with low methanogen abundance in water based on metagenomic analysis, N₂ injection did not increase CH₄ production (Fig. 3a, b).

Site-scale investigation of N₂O productions in sediments vary with trophic status

Nitrification and denitrification represent the primary N₂O production pathways through NO reduction carried out by nitric oxide reductase (NOR)^21,22. AOB additionally produce N₂O via cytochrome P460 (CytL) during nitrification²³. The non-random co-occurrence of denitrifying genes shows that nosZ associates more frequently with nirS than nirK, making nirS-type denitrifiers weaker N₂O emitters than nirK-type through complete denitrification⁴⁹. Soil N₂O sink capacity correlates strongly with the nirS/nirK ratio, where higher ratios indicate greater N₂O sink capacity⁵⁰. In sediments of Lake DH, lower nirK abundance relative to nirS (Figs. S5a, S9) suggests efficient N₂O consumption potential through complete denitrification. This aligns with the previous observation that eutrophic lake sediments favor nirS-type denitrifiers capable of complete denitrification, while oligotrophic systems harbor more nirK-type denitrifiers⁵¹. In contrast, water column gene proportions (low nirS/nirK ratio) indicate greater denitrification-derived N₂O emission potential (Fig. S5a), likely due to the low organic carbon concentrations (Table S1). Given the weak abundance of nitrifying and denitrifying genes in the water column (Fig. S7), which predicts much lower nitrogen transformation and N₂O emission potentials than in sediments.

Microbial N₂O reduction to N₂, mediated by NosZ, governs biological N₂O sinks. Phylogenetic analysis of nosZ (Fig. S10) identified clades I and II⁵², with clade II more abundant in both sediment and water (Figs. S5a, S10). Interestingly, nosZ richness was substantially higher in nirS-dominated sediments than nirK-dominated water (Fig. S10). This pattern matches a previous observation in soils, where nirK dominance correlates with low nosZ diversity, further structural equation model suggested that clade II nosZ abundance and diversity exhibit positive correlations with N₂O sink capacity (consume more N₂O than it produces)⁵⁰. Consistent with the high nirS/nirK ratio, sediments exhibited significantly greater abundance and phylogenetic diversity of clade II nosZ genes compared to the water column (Figs. S5a, S10), indicating enhanced potential for N₂O sink in sediments.

To distinguish N₂O production from nitrification and denitrification, we employed dicyandiamide (DCD), a specific ammonia oxidation inhibitor⁵³, to selectively block nitrification. Following 12-hour incubation with 2 mM DCD, sediment nitrification in Lake DH was completely suppressed while denitrification remained unaffected (Fig. 4a). Control experiments (without DCD) showed substantial sedimentary N₂O emissions (6.33 ± 0.30 nmol N₂O/g_(sediment)/h) but minimal production from water columns (Fig. 4b). Notably, DCD treatment reduced sedimentary N₂O emissions to negligible levels (Fig. 4b), demonstrating that denitrification contributed minimally to N₂O production. Because DCD specifically intercepted the first step of ammonia oxidation that the following N₂O production pathways in nitrification, including NH₂OH oxidation and nitrifier denitrification^21,23,24, were completely blocked, leaving denitrification as the sole N₂O source. Conversely, despite relatively low abundances of N₂O-production genes (NorB and CytL, Fig. 1), nitrification accounted for the majority of N₂O emissions in these eutrophic sediments. The nitrification-dominant N₂O production pattern was replicated in sediments of another eutrophic Lake Kangle (KL) (Fig. 4c), which were also rich in organic carbon (total organic carbon (TOC), 24.86 mg/g) and inorganic nitrogen (NH₄⁺, 12.76; NO₂⁻, 0; NO₃⁻, 9.8 mg-N/kg) (Table 1). In contrast, DCD failed to suppress the sedimentary N₂O emissions in oligotrophic Lake Ranwu (RW) (low TOC, 0.32 mg/g; NH₄⁺, 0; NO₂⁻, 0; and NO₃⁻, 0.42 mg-N/kg) (Table 1), confirming incomplete denitrification rather than nitrification as the primary driver of N₂O production in oligotrophic sediments. Expectedly, eutrophic lake sediments exhibited markedly higher N₂O production rates via nitrification compared to oligotrophic sediments where denitrification dominated N₂O production (Fig. 4b, c).

Above findings align with established denitrification dynamics in oceans that complete denitrification (NO₃^– → N₂) predominates organic matter-rich environments (high organic matter:NO₃^– ratio), whereas low organic availability favors incomplete denitrification (NO₃^– → N₂O) and consequent N₂O accumulation⁵⁴. Supporting this, wastewater treatment systems similarly show denitrification typically functions as a net N₂O sink, with reduction rates exceeding production by 2–10 fold⁵⁵. The high sedimentary TOC:NO₃^– ratios in Lake DH (average 0.7) and KL (0.6) likely stimulated complete denitrification, resulting in negligible denitrification-derived N₂O emissions (Fig. 4b). In contrast, the oligotrophic Lake RW exhibited much lower ratio (0.3), suggesting a limited organic carbon supply likely led to incomplete denitrification and higher N₂O yield.

Global patterns of sediment CH₄ and N₂O emissions linked to trophic status

Methanogenic archaea, the sole organisms producing CH₄ as part of their energy-generating metabolism, including the only archaeal diazotrophs known to contain nitrogenase gene (nifH) for nitrogen fixation³⁸. Supporting our batch cultivation results (Fig. 3c, d), global metagenomic analysis revealed a strong correlation between archaeal mcrA and nifH (p < 0.001; Fig. 5a), indicating widespread methanogenesis-nitrogen fixation coupling in lake sediments. However, mcrA (p > 0.05) and archaeal nifH (p > 0.05) abundances did not vary with trophic state (Fig. 5b), suggesting methanogenesis potential in lake sediments with varying eutrophic levels is multi-factorial and unpredictable. CH₄ production rate by methanogenic archaea in lake sediments can be significantly affected by temperature⁵⁶, concentrations of iron and sulfur⁵⁷, lake depth⁵⁸, lake size and productivity⁵⁹, and sediment accumulation⁶⁰. Consequently, the lack of worldwide data on these drivers has limited their incorporation into models for estimating CH₄ emissions from lakes¹⁶. In contrast, the abundance of bacterial nifH gene is significantly affected by trophic status (p < 0.001), with nitrogen-fixing bacteria showing a preference for lakes with high eutrophic conditions (Fig. 5b). Our result aligns with enhanced cyanobacterial proliferation under nutrient-enriched conditions, as cyanobacteria possess efficient metabolic pathways for utilizing organic phosphorus compounds⁶¹. In addition, high nitrogen fixation rates were measured in nearshore waters with high dissolved inorganic nitrogen concentrations^62,63, even though elevated dissolved inorganic nitrogen has historically been thought to inhibit nitrogen fixation⁶⁴. These results suggest that high concentration of fixed nitrogen does not necessarily suppress bacterial nitrogen fixation.

Our global analysis revealed distinct niche partitioning between nirK– and nirS-type denitrifiers across lake sediments with distinct trophic levels, evidenced by significant variation in nirS/nirK ratios with trophic status (p < 0.001; Fig. 5c). A strong N₂O sink capacity was predicted in eutrophic lakes and indicated by the significantly higher nirS abundance and nirS/nirK ratios (p < 0.01; Fig. 5c). This pattern extended to norB and nosZ genes related to N₂O generation and reduction, which showed parallel distribution trends with nirS (Fig. 5c), suggesting that eutrophic systems maintain high genetic potential for complete denitrification pathways. The elevated organic carbon availability in these nutrient-rich environments likely provides sufficient electron donors to support robust N₂O reduction to N₂, consistent with the neglectable denitrification-derived N₂O production rates we observed in eutrophic sediments of Lake DH (Fig. 4b) and Lake KL (Fig. 4c). In contrast, oligotrophic lakes exhibited fundamentally different N₂O dynamics, with significantly higher nirK abundance and lower nirS/nirK ratios (p < 0.001; Fig. 5c) indicating greater N₂O emission potential via incomplete denitrification. This prediction was confirmed by subsequent batch cultivation experiments using oligotrophic sediments from Lake RW (Fig. 4c). In these low-nutrient systems, electron competition during upstream denitrification steps appears to constrain N₂O reduction activity⁶⁵, resulting in substantial N₂O accumulation.

Cross-inoculation demonstrates trophic regulation of N₂O production

Eutrophication due to nutrient enrichment can significantly increases N₂O emissions from shallow lakes⁶⁶, while enhanced sedimentary N₂O production has been primarily attributed to denitrification^67,68,69, the role of nitrification is usually overlooked. In this study, oligotrophic RW sediments were inoculated into sterilized eutrophic KL sediments at proportions of 25%, 50%, and 75% to simulate eutrophication (Table 1, Fig. 6g). Conversely, KL sediments were inoculated into sterilized RW sediments at the same proportions to simulate oligotrophication process (Table 1, Fig. 6g).

In oligotrophic RW sediments, N₂O emissions were primarily driven by denitrification (Fig. 4c). Similarly, at mild eutrophic (KL:RW = 1:1) and mesotrophic (KL:RW = 1:3) conditions, the addition of DCD (nitrification inhibitor) did not significantly reduce sedimentary N₂O emissions, confirming denitrification remained the dominant N₂O production pathway (Fig. 6d, f). However, at moderate eutrophic (KL:RW = 3:1) level, DCD addition strongly suppressed N₂O emission (Fig. 6b), indicating nitrification progressively took over as the main N₂O production pathway as eutrophication advanced. In line with this finding, nitrification was identified as the main source of N₂O in eutrophic Lake Taihu²⁸. As it was found that the produced N₂O by denitrification was simultaneously eliminated via N₂O reduction²⁸, a process likely driven by high concentrations of dissolved organic carbon (DOC), which acted as an electron donor for complete denitrification. This stimulatory effect of DOC on N₂O consumption is consistent with observations in diverse island waters⁷⁰. Conversely, in eutrophic KL sediments, nitrification initially dominated N₂O production (Fig. 4c). Similarly, at moderate eutrophic condition, N₂O production was primarily attributed to nitrification, as evidenced by the strong suppression of N₂O emissions upon the addition of DCD (Fig. 6a). However, when oligotrophication exceeded mesotrophic level, nitrification inhibition (via DCD) no longer substantially reduced N₂O emissions (Fig. 6), demonstrating a gradual shift to denitrification as the predominant pathway. These results reveal a reversible trophic status-dependent transition in N₂O emission pathways: from denitrification- to nitrification-dominance during eutrophication, and the inverse during oligotrophication. This underscores trophic status as a critical regulator of N₂O dynamics in lake sediments. Consistently, the measured nirS/nirK ratio increased sharply with elevated eutrophic status (Fig. S11), indicating a systematic shift toward more complete denitrification and reduced N₂O production. These results strongly support the hypothesis that the nirS/nirK ratio is a reliable indicator of denitrification completeness and its associated N₂O production potential. Aligns with our global metagenomic analysis on methanogenesis potential in lake sediments (Fig. 5b), there was no significant (p > 0.05) correlation between CH₄ production rate and the eutrophication level in both eutrophication and oligotrophication processes (Fig. S12). While this study specifically examined freshwater lakes, the proposed theory may potentially extend to other unexamined aquatic and terrestrial ecosystems, including saltwater lakes, rivers, wetlands, soils, and marine systems.

Our study uncovers a critical trophic status-regulated linkage between CH₄ and N₂O emissions in freshwater lake sediments (Fig. 7). We show that methanogenesis is tightly coupled with nitrogen fixation, while exhibiting complex and context-dependent responses to eutrophication. Crucially, trophic status emerges as a key determinant of N₂O production sources. In eutrophic lakes, enhanced nitrogen removal via coupled nitrification-denitrification drives substantial N₂O emissions, with nitrification serving as the dominant production pathway. In contrast, oligotrophic lakes predominantly generate N₂O through incomplete denitrification, exhibiting lower overall emission potentials. Consequently, as global lake eutrophication is projected to intensify in the coming decades under a fossil-fueled development pathway³³, is expected to increase the lacustrine N₂O production through nitrification stimulation, while marginalizing denitrification-derived fluxes. To counteract this, precise ammonium reduction management strategies, such as controlling external inputs, managing endogenous pollution, and enhancing nitrogen removal via anammox, could effectively mitigate N₂O emissions. Conversely, sustainable nutrient management may trigger the oligotrophication of eutrophic lakes, progressively shifting N₂O production toward denitrification-dominated regimes. Effective management would therefore involve strategies such as scavenging accumulated nitrate from sediments or increasing the C/N ratio by replenishing them with nitrogen-free organic carbon to restrict incomplete denitrification.

However, limitations are existing in our work that laboratorial batch experiments cannot fully capture in situ lake conditions, such as sediment-water interactions, seasonal variability. Future research should concentrate on: (ⅰ) validating the crucial findings generated in this study using, such as mesocosms or long-term monitoring experiments, (ⅱ) elucidating how nitrogen cycling modulates GHG fluxes under combined global warming and eutrophication scenarios, (ⅲ) developing targeted management strategies to control GHG emissions while maintaining ecosystem function, and (ⅳ) incorporating more global data on the drivers of CH₄ production is essential to constrain model uncertainties regarding its response to lake eutrophication. Such advances will be essential for informing climate-smart water quality policies and achieving meaningful GHG reductions from inland waters.

Study site and sampling protocol

Lake DH (30.55°N, 114.38°E), located in Wuhan, China (Fig. S1). The lake map was modified from our previous study⁷¹, which was initially produced by ArcGIS v10.7. As the second largest urban lake in China, this shallow subtropical waterbody (mean depth: 2.2 m; maximum depth: 4.8 m) exhibits characteristic features of long-term eutrophication, including elevated greenhouse gas emissions (CH₄: 23.3 ± 18.6 mg/m²_(water)/d; N₂O: 40.32 mg N/kg_(sediment)/d)^72,73. In the initial sampling campaign on August 1^st, 2020, we collected 25 sediment samples (0–20 cm) from 5 representative sites (D1–D5; n = 5 per site) using gravity sampler. Sediment samples were collected from 0–20 cm in this study, the depth layer with the most active microbial metabolism. We also obtained 15 water column samples (1 m depth) from sites D1–D3 and D6–D7 (n = 3 per site), samples were then mixed and filtered by a 0.22 μm filter membrane to capture microbial biomass. Samples from D6–D7 (sediment) and D4–D5 (water) were compromised during transport and excluded from analysis. Samples were immediately transported to the laboratory in dry ice and stored at −80 °C until total DNA extraction to preserve microbial community integrity.

In the follow-up sampling on September 26^th, 2023, triplicate sediment and water samples were collected from sites D1–D3 using above sampling procedures for DNA extraction. Fresh samples were also collected from sites D1–D3 for batch cultivation and isotopic tracing experiments. Sediment samples (0–20 cm) were collected using a gravity sampler, while water samples were collected at the depth of 1 m, 2 m, and 3 m below the water surface (adjusted to 1 m and 1.8 m at site D3). Samples were transported back to the lab using ice bag and stored at 4 °C. Field replicates were homogenized prior to laboratory analyses to ensure representative sampling subsequent. Detailed protocols for biogeochemical parameters measurements, DNA extraction and sequencing, and metagenomic data analysis are provided in Supplementary Information. While corresponding water samples from sites D6–D7 and sediment samples from sites D4–D5 were not successfully collected or sequenced, the overall nitrogen cycling processes at sites D1–D3 are highly similar (Fig. 1), indicating a high degree of consistency across different sites in Lake DH. Therefore, only samples collected from D1–D3 were included for the final analysis in Fig. 1.

To simulate the eutrophication and oligotrophication process by sediment cross-inoculation experiments, we collected fresh sediment samples (0–20 cm) from an oligotrophic Lake RW (29.45 N, 96.80E) on the Qinghai-Tibet Plateau on April 20^th, 2025 and a eutrophic Lake KL (23.09 N, 113.29E) on Sun Yat-sen campus on April 16^th, 2025. Biogeochemical parameters of these two lakes were measured and trophic state indexes were calculated (Tables 1, S3).

Batch cultivation on N₂O and CH₄ production

Dicyandiamide (DCD) (99.7%, Shanghai Macklin Biochemical, Shanghai, China) is a typical nitrifying inhibitor⁵³, which was used to distinguish the N₂O emission rates of nitrification and denitrification by specifically inhibiting the activity of nitrification. First, we investigate the inhibitory effect of DCD on nitrification, a control group (0.5 mM NH₄⁺, (≥99.5%, Sigma-Aldrich, USA)) and an experiment group (0.5 mM NH₄⁺ and 2 mM DCD) were designed, and each had three biological replicates. Furthermore, to exclude the effect of DCD on denitrifying activities in sediments, a control group (0.5 mM NO₃⁻, 99.99%, Shanghai Aladdin Biochemical Technology, Shanghai, Chin) and an experiment group (0.5 mM NO₃⁻ and 2 mM DCD) were designed. We inoculated 10 mL of fresh sediment and water samples into sterilized 50 mL serum bottles, which were securely sealed with butyl rubber septa. These bottles were incubated at 30 °C with a shaking speed of 100 rpm for 48 h. The concentrations of NH₄⁺ were measured photometrically⁷⁴ and NO₃⁻ was reduced to NO₂⁻ by vanadium chloride and measured photometrically⁷⁵ at intervals of 0, 6, 12, 24, and 48 h.

To investigate the N₂O production rates of nitrification and denitrification under the low dissolved oxygen (DO) condition in lake sediments (< 3 mg/kg), two-treatment experiments were designed for both water and sediment samples: (i) 200 μM NH₄⁺ and 200 μM NO₃⁻, (ii) 200 μM NH₄⁺, 200 μM NO₃⁻, and 2 mM DCD, and each had three biological replicates. First, we inoculated 10 mL of either fresh sediment or water into sterilized 50 mL serum bottles and half of them were pretreated with 2 mM DCD, which were pre-incubated at 28 °C with a shaking speed of 100 rpm for 12 h to make sure that nitrification activity was totally inhibited by DCD. Ar was pumped into each bottle to create a low DO condition (DO < 3 mg/L). After that, 200 μM NH₄⁺ and 200 μM NO₃⁻ were injected into each bottle, and then incubated at 28 °C with a shaking speed of 100 rpm. Gas samples from the headspace were collected at intervals of 0, 6, 12, 24, and 48 h, and the concentration of N₂O was analyzed using a gas chromatograph (Fuli GC9790Plus).

Batch cultivation on CH₄ production

To identify the coupling between methanogenesis and N fixation sediment and water samples, a control group using argon (Ar, 99.99%, Shenzhen Huatepeng SPECIAL Gases, Shenzhen, China) as the headspace and an experiment group using N₂ (99.999%; Shenzhen Huatepeng SPECIAL Gases, Shenzhen, China) as the headspace were designed, and each had three biological replicates. We inoculated 10 mL fresh sediment or water into each sterilized 50 mL serum bottle, which were securely sealed with butyl rubber septa, and sextuple samples for both water and sediment were prepared. The control group was purged by Ar for 30 minutes to squeeze out air and maintain an anaerobic state, and the experiment group was purged by N₂ for 30 minutes. These treatments were incubated at 30 °C with a shaking speed of 100 rpm. Gas samples from the headspace were collected at intervals of 0, 6, 12, 24, and 48 h and the concentration of CH₄ was analyzed using a gas chromatograph (Fuli GC9790Plus), and the concentration of NH₄⁺ in sediments was measured photometrically.

Sediment cross-inoculation experiments

To investigate the effect of trophic status on sediment N₂O production dynamics, a cross-inoculation experiments of oligotrophic (Lake RW) and eutrophic (Lake KL) sediments was designed to simulate eutrophication and oligotrophication processes. About half of these sediment samples were sterilized under 121 °C for 30 min by a high-pressure steam sterilizer to inactivate all microorganisms. To simulate the eutrophication process, fresh Lake RW sediment samples were inoculated into the sterilized Lake KL sediments with the proportions of 25, 50, and 75%. In contrast, fresh Lake KL sediment samples were inoculated into the sterilized Lake RW sediments with the proportions of 25, 50, and 75% to simulate the oligotrophication process. The concentrations of DOC, NH₄⁺, and NO₃⁻ in composite sediment samples were measured, which were then statically incubated in the dark at 28 °C for one month. Throughout the incubation process, concentrations of DOC, NH₄⁺, and NO₃⁻ were measured every week and replenished to initial concentrations. Samples from these composite sediments were collected every week for N₂O production batch experiments. For each experiment, 10 g sample was mixed with 10 mL of ultrapure water in a 50 mL serum bottle, followed by two-treatment experiments (with or without DCD) as described above. Prior to sealing the bottle, the headspace of the bottle was purged with clean air, filtered by 0.22 μm filter membrane. Six air samples were then collected during this purging process and designated as the time 0 samples for this batch, to ensure that all experiments in this batch started under the same initial conditions. Gas samples from the headspace were collected at intervals of 0, 6, 12, 24, and 48 h and the concentration of N₂O was analyzed using a gas chromatograph (Agilent GC 8890). To detect the production of CH₄, 10 g sediment sample was mixed with 10 mL of ultrapure water in a 50 mL serum bottle followed by incubation experiments as described above. Gas samples were collected daily from the headspace for a total of seven days, and the concentration of CH₄ was analyzed using a gas chromatograph (Agilent GC 8890).

qPCR analysis of nirS and nirK gene abundance

After one month of cultivation, DNA was extracted from the sediment samples. The abundances of the nirS and nirK were quantified by qPCR using a CFX Opus 96 Real-Time PCR System (Bio-Rad Laboratories, CA, USA). NirK gene-specific primers (5’–3’), nirK876 (ATYGGCGGVAYGGCGA) and nirK1040 (GCCTCGATCAGRTTRTGGTT)^76,77, were selected. The protocol consisted of an initial denaturation at 95 °C for 15 min, followed by 40 cycles of 95 °C for 20 s, 60 °C for 35 s, and 72 °C for 70 s. NirS gene primers (5’–3’), nirSCd3aF (AACGYSAAGGARACSGG) and nirSR3cd (GASTTCGGRTGSGTCTTSAYGAA)^77,78, were selected. The protocol consisted of an initial denaturation at 95 °C for 15 min, followed by 40 cycles of 95 °C for 20 s, 55 °C for 35 s, and 72 °C for 70 s. Reactions were carried out in full-skirted 96-well plates (cat. no. PC-96-FS-0100-WC, Ranjeck Technology, Beijing, China) with sealing film (cat. no. SF-001-UC-100B, Ranjeck Technology, Beijing, China). All reactions were performed in a final volume of 20 μL, containing 1 μL of template DNA, 10 μL of 2 ⅹ iTaq Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA), 0.8 μL of each forward and reverse primer (10 μM each), and 8.2 μL of RNase-free water. Standard curves for absolute quantification were generated for each assay using serial dilutions of plasmids containing the respective nirK or nirS gene, which were constructed via TA cloning. The target fragment was amplified by PCR. The PCR product was then purified by gel extraction using TaKaRa MiniBEST DNA Fragment Purification Kit Ver.4.0 (Takara Bio, Shiga, Japan). TA cloning was performed using the Mighty TA-Cloning Kit (Takara Bio, Shiga, Japan). The recombinant plasmid was introduced into DH5α Super Competent Cells (Shanghai Biyuntian Biological, Shanghai, China) and cultured in LB medium containing ampicillin. Following incubation, colony PCR was carried out to identify positive clones. The positive colonies were then selected for plasmid extraction using the TaKaRa MiniBEST Plasmid Purification Kit Ver. 4.0 (Takara Bio, Shiga, Japan).

¹⁵N isotope traces nitrogen removal rates

Following the sample preparation procedure described above, ¹⁵N isotope was used to trace the nitrogen removal rates by anammox and denitrification. We inoculated 4 mL of either fresh sediment or water into sterilized 12 mL vials with three biological replicates, which were then pre-incubated at 30 °C and a shaking speed of 100 rpm until all NH₄⁺, NO₂⁻ and NO₃⁻ were consumed to undetectable concentrations. A final concentration of 2 mM DCD was injected to totally inhibit nitrification process with a shaking speed of 100 rpm for 12 h. After that, 200 μM ¹⁴NH₄⁺ and 200 μM ¹⁵NO₃⁻ (99.14 atm%, Shanghai Research Institute of Chemical Industry, Shanghai, China) were injected into these preprocessed samples, and each sample was purged by Ar for 30 min to squeeze out air and maintain an anaerobic state. Samples were incubated at 30 °C with a shaking speed of 100 rpm. The incubations were stopped at defined intervals (0, 6, 12, 24, and 48 h) by adding 200 μL of saturated ZnCl₂. The concentrations of ²⁹N₂ to ³⁰N₂ in the head space of the vials were measured by the Third Institute of Oceanography of the Ministry of Natural Resources, PRC.

Analysis of global metagenomic data of freshwater lake sediments

Our in-house metagenomic data were complemented by available public datasets from freshwater lake sediments, which were retrieved from the NCBI SRA on or before May 30, 2024^{79,80,81,82,83,84,85,86,87,88,89,90,91}. A total of 77 metagenomic datasets were collected worldwide, including 40 from eutrophic and 37 from oligotrophic lakes. These samples spanned a wide geographic and latitudinal range, from the tropics to the Arctic, across Asia, Europe, and North America. This broad coverage enhances the generalizability of our global metagenomic analysis. It was conducted as described in Supplementary Information, and the assembled nitrogen and methane cycle-related genes were identified. Sequences of nirS and nirK genes originating from nitrifiers and anammox bacteria were identified within the denitrification gene pool using BLASTP (identity > 70%) against constructed custom databases for AOB and anammox bacteria⁹². These non-denitrifier sequences were subsequently excluded from the nirS/nirK ratio calculation. All the assemblies were merged together and the coverage of the recovered gene in each sediment sample was calculated by mapping the clean reads using ‘coverm contig’ modules of CoverM v0.6.1. To construct the correlation between the abundance of mcrA and archaeal nifH, a linear regression model was employed to calculate the regression curve, accompanied by the presentation of adjusted R-squared value and p-value. Boxplots were used to compare the abundance of CH₄ and N₂O emission related genes in lakes with different trophic status. The boxplots are generated using the R package ggpubr. Outliers produced by the extremely high abundance of some genes were deleted based on Tukey’s test and z-score normalization prior to plotting (less than five per box). Points with zero values in either the numerator or the denominator are eliminated prior to calculating the ratios of different genes. The global map was produced by R package rnaturalearth.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.